deadbydawn101

Intermediate
@DeadByDawn101

Building @RavenXllm @jolyneshibasol

turboquant-mlx. First MLX implementation of TurboQuant KV cache compression for Apple Silicon

29

OpenMythos-MLX. OpenMythos ported to Apple Silicon MLX — Recurrent-Depth Transformer with MoE, MLA, ACT halting

12

RavenX-Sec. Autonomous Security Intelligence Model — find, classify, remediate, and report vulnerabilities in real-time

6

ravenx-memory. Hybrid triple-backend agent memory — Markdown source of truth + SQLite FTS5 + QMD vectors. Harness-agnostic (Hermes + OpenClaw), model-agnostic, MCP-native. Forked from Icarus, stripped of NVIDIA/OpenAI dependencies, ported to MLX.

6

OpenSelfRevise. Open-source implementation of Self-Revising Discovery Systems (MIT, arXiv:2606.01444). Typed artifact systems with MDL-gated regime transitions, Builder/Breaker adversarial self-revision, and Kan-extension knowledge transport. Trainable method for any domain.

3

made-in-heaven. Autonomous web agent powered by Chrome DevTools MCP — full browser control, zero automation fingerprint, persistent auth

2

tesla-grok-road-agent. OpenClaw + Grok-4 agent that turns Tesla FSD road brainstorms into shipped code + live X posts. Ideas born on the road, posted before you park.

1

star-platinum-cluster. Python

1

ravenx-agent-safety. Agent safety mitigations from Claude Mythos Preview system card

1

ravenx-inference-harness. RavenX inference harness — mlx-lm + TriAttention + TurboQuant on Apple Silicon

1

OpenDiffusion-MLX. DiffusionGemma ported to MLX — 4x faster text generation via diffusion on Apple Silicon. 256-token parallel canvas, bidirectional self-correction, block autoregressive.

1

ravenx-mtp-drafter. Multi-Token Prediction drafter for speculative decoding — model-agnostic, MLX-native. Extracted architecture from Google's hidden Gemma 4 MTP heads. Train custom drafters for ANY model on Apple Silicon.

1

RavenX-Trade. Autonomous trading intelligence model. Predict trades, quants, memecoins, stocks. Built by RavenX LLC on Apple Silicon.

1

OpenFable-MLX. Identity-persistent AI model training framework for Apple Silicon — MLX + Metal. Train models that know who they are. Part of the RavenX open-source AI stack.

1
14
Apply